CONUT score and Naples prognostic score (NPS) for dual prediction of survival and radiation toxicity in locally advanced non-small cell lung cancer with chemoradiotherapy Authors: Shuyang Chen, Xin Qu, Jingyi Wang, Zhongyu Li, Yang Cao, Chunbo Wang Published: 2026 ABSTRACT BackgroundAlthough chemoradiotherapy (CRT) is the standard modality for the treatment of locally advanced non-small cell lung cancer (LA-NSCLC), there have been inconsistencies in the efficacy and toxicity of these treatment regimens. To address this issue, we sought to evaluate peripheral blood-derived nutrition-inflammation scoring indices as dual-purpose tools for predicting both survival outcomes and treatment-related toxicities in this patient population.MethodsThe study included patients with LA-NSCLC who underwent CRT at Harbin Medical University Cancer Hospital between March 2019 and December 2022. Peripheral blood indicators before radiotherapy were used to calculate the neutrophil-to-lymphocyte ratio (NLR), prognostic nutritional index (PNI), advanced lung cancer inflammation index (ALI), controlled nutritional status score (CONUT score), and Naples prognostic score (NPS). We analyzed and compared their relationships with prognosis and radiotherapy-related toxicity events.ResultsIn total, 126 patients were included in the study, with a median follow-up period of 53 months. In terms of survival prediction, the multivariate Cox regression analysis revealed that a high CONUT score [progression-free survival (PFS): hazard ratio (HR) =2.513, 95% confidence interval (CI): 1.336–4.727, P=0.004; overall survival (OS): HR =2.908, 95% CI: 1.527–5.538, P=0.001) and high NPS (PFS: HR =2.809, 95% CI: 1.517–5.203, P=0.001; OS: HR =3.345, 95%CI: 1.769–6.325, P<0.001) were independent risk factors for PFS and OS. Receiver operating characteristic (ROC) curve analysis indicated that when predicting 24- and 36-month PFS, the area under the curve (AUC) values of CONUT score were 0.791 (95% CI: 0.704–0.878) and 0.800 (95% CI: 0.724–0.877), respectively, and the corresponding values for the NPS were 0.741 (95% CI: 0.657–0.825) and 0.790 (95% CI: 0.715–0.866). When predicting 3- and 5-year OS, the AUCs of CONUT scores were 0.823 (95% CI: 0.746–0.900) and 0.817 (95% CI: 0.745–0.889), respectively, and the AUCs of NPS were both 0.802 (95% CI: 0.728–0.877, 95% CI: 0.727–0.878). In terms of predicting radiotherapy toxicity, grade ≥2 radiation pneumonitis or grade ≥2 radiation esophagitis served as the observation endpoints for assessing radiotherapy toxicity. Multivariable logistic regression analysis revealed that a high CONUT score [odds ratio (OR) =3.194, 95% CI: 1.001–10.207, P=0.049] and NPS group 2 (OR =8.243, 95% CI: 2.946–23.069, P<0.001) were associated with radiotherapy toxicity endpoint events.ConclusionsAmong LA-NSCLC patients receiving CRT, the CONUT score and NPS prior to treatment were established to be associated with survival and grade ≥2 radiation-induced lung/esophageal toxicity. Prospective studies are required for further verification to guide clinical practice. Highlight box Key findings • The controlled nutritional status score (CONUT score) and Naples prognostic score (NPS) may provide a simple stratification tool covering both survival prognosis and toxicity risk for patients with locally advanced non-small cell lung cancer (LA-NSCLC) receiving chemoradiotherapy (CRT). What is known and what is new? • The nutritional and inflammatory status of lung cancer patients is significantly associated with their survival prognosis. • The CONUT score and NPS not only independently predict survival but are also correlated with radiotherapy toxicity. What is the implication, and what should change now? • The CONUT score and NPS are economical and accessible bedside risk-stratification tools that can effectively identify high-risk populations and provide preventive nutritional support. • To enhance clinical value, prospective intervention studies based on these scores should be conducted. Introduction Lung cancer is a malignant tumor with high morbidity and mortality rates. Among its histological subtypes, non-small cell lung cancer (NSCLC) accounts for approximately 85%, and 30–40% of patients are diagnosed with stage III NSCLC (1,2). Owing to extensive lymph node involvement or significant comorbidities, the majority of stage III NSCLC cases, particularly stages IIIB and IIIC, are deemed ineligible for surgical resection (3). Studies such as West Japan Lung Cancer Group (WJLCG) and Radiation Therapy Oncology Group (RTOG) 0617 have established chemoradiotherapy (CRT) as the cornerstone of treatment for patients with unresectable locally advanced non-small cell lung cancer (LA-NSCLC) (4,5). Although the PACIFIC, GEMSTONE-301, LAURA, and POLESTAR studies improved treatment efficacy through consolidation immunotherapy or targeted maintenance therapy, CRT remains the foundation of subsequent treatment regimens. Despite these advancements, substantial inter-patient heterogeneity leads to variable clinical outcomes, posing a formidable challenge to the current “one-size-fits-all” approach (6,7). Furthermore, radiation toxicity, including severe fatigue, esophagitis, and pneumonitis, not only compromises the quality of life and treatment compliance but also precipitates treatment interruptions that ultimately erode survival benefits (7). Adverse reactions during CRT, such as fatigue, radiation esophagitis, and radiation pneumonitis, can reduce the quality of life of patients, influence treatment compliance, and even lead to treatment interruption, thereby influencing survival prognosis. Consequently, from the perspectives of enhancing the monitoring of toxic events or for considering adjuvant treatment strategies, identifying patients with a poor prognosis is of considerable importance in current clinical practice. Accumulating evidence underscores the close interplay between nutritional status, systemic inflammation, and malignant tumorigenesis. Host nutritional status modulates the functional repertoire of immune cells, thereby orchestrating the antitumor immune response and influencing cancer progression. Chronic inflammation remodels the tumor immune microenvironment and activates oncogenic signaling pathways, thereby fostering cellular proliferation and metastatic dissemination (8,9). Reflecting this biology, peripheral blood biomarkers, such as the neutrophil-to-lymphocyte ratio (NLR), prognostic nutritional index (PNI), and advanced lung cancer inflammation index (ALI), have emerged as promising prognosticators in NSCLC, offering a minimally invasive snapshot of the host status (10-13). However, these unidimensional metrics often fail to capture the complex dynamic interplay between nutritional deprivation and systemic inflammation. A single nutritional or inflammatory biomarker remains insufficient to comprehensively reflect patients’ overall condition, treatment response, and prognosis. There is an urgent need to establish a clinical risk prediction model that combines multiple indicators, such as host nutritional status and inflammation-related indicators. Controlled nutritional status score (CONUT score) is a scoring system that reflects nutritional status by assigning values to absolute lymphocyte count, albumin (ALB), and total cholesterol (TC) levels. It has been proven to independently predict the survival benefits of lung cancer patients (14). The Naples prognostic score (NPS) is a novel prognostic scoring system based on ALB, TC, NLR and lymphocyte-to-monocyte ratio (LMR). Previous studies have confirmed its prognostic value in evaluating patients with unresectable LA-NSCLC (15). However, at present, there is no systematic study comparing the prognostic predictive performance of the above-mentioned models in patients with LA-NSCLC receiving CRT. Meanwhile, there is no study evaluating the prediction of radiotherapy toxicity by clinical predictive models covering nutritional and inflammatory status markers. Therefore, this study retrospectively collected data from patients with IIIB–IIIC NSCLC who received CRT, aiming to find a promising nutrition-inflammation scoring model that can predict the survival prognosis and radiotherapy toxicity of patients with LA-NSCLC who received CRT. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1466/rc). Methods Research population This study reviewed patients with NSCLC who received CRT and were admitted to Harbin Medical University Cancer Hospital from March 1, 2019 to December 30, 2022. This study included patients with stage IIIB/IIIC NSCLC who received CRT. Patients with stage IIIA were not included as they routinely received surgical treatment. The inclusion criteria were as follows: (I) histologically confirmed diagnosis; (II) TNM stage IIIB–IIIC defined by the 8th edition of the International Association for the Study of Lung Cancer (IASLC); (III) at least one measurable lesion; (IV) Eastern Cooperative Oncology Group performance status (ECOG PS): 0–1 point; (V) no previous history of treatment for other tumors. The exclusion criteria were as follows: (I) hematological or immune system diseases; (II) history of cardiac, hepatic, or renal dysfunction; (III) missing key laboratory data; (IV) loss to follow-up or follow-up <3 months. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Harbin Medical University Cancer Hospital (No. KY2021-01). Individual consent for this retrospective analysis was waived. Treatment plan The treatment regimen included 6 weeks of intensity-modulated radiation therapy (IMRT) combined with platinum-based chemotherapy. The radiotherapy dose is 2.0 Gy once daily, five times a week, with a total dose of 60 Gy. The target area for radiotherapy was determined by combining the imaging results of positron emission tomography/computed tomography (PET/CT) and enhanced chest CT, and involved-field irradiation was used. Chemotherapy was administered every 21 days for 4 cycles. Of these patients, 94 received the cisplatin (75 mg/m2)/paclitaxel (135 mg/m2) regimen and 32 patients received the cisplatin (75 mg/m2)/pemetrexed (500 mg/m2) regimen. The complete patient selection workflow and grouping schema are illustrated in Figure 1. As of the follow-up cut-off date, having undergone CRT, none of the patients in this retrospective cohort had received consolidation immunotherapy or targeted maintenance therapy. Flow diagram. ECOG PS, Eastern Cooperative Oncology Group performance status; IASLC, International Association for the Study of Lung Cancer; IMRT, intensity-modulated radiation therapy; LA-NSCLC, locally advanced non-small cell lung cancer; TNM, tumor-node-metastasis. Evaluation of baseline NLR, PNI, ALI, CONUT score and NPS Laboratory test results obtained within one week prior to radiotherapy were utilized. The NLR was calculated as the ratio of the absolute neutrophil count to the absolute lymphocyte count. The PNI was determined using the formula: ALB (g/L) + 5 × Absolute lymphocyte count (109/L). The ALI was calculated as BMI × ALB (g/L) × 0.1/NLR. The CONUT score is presented in Figure 2. The NPS is depicted in Figure 3. The cutoff values employed in this study were derived from previous literature, specifically NLR 5 (16,17), PNI 45 (18), ALI 18 (19), and CONUT score 4 (14). The sensitivity analysis results are detailed in Figure S1 and Table S1. The NPS was categorized into group 0, group 1, and group 2, as illustrated in Figure 3. Tumor response was assessed in accordance with the Response Evaluation Criteria for Solid Tumors (RECIST) version 1.1 (20). Additionally, radiotherapy-related toxicities were evaluated in accordance with Version 4.0 of the Common Terminology Criteria for Adverse Events (CTCAE-v4.0) (21). CONUT score. CONUT score, controlled nutritional status score. Naples prognostic score. NPS, Naples prognostic score. Follow-up Follow-up assessments were conducted through telephone calls, text messages, or WeChat. Progression-free survival (PFS) is defined as the duration from the initiation of radiotherapy to the date of disease progression, while overall survival (OS) is defined as the duration from the initiation of radiotherapy to death from any cause. The primary tumor and distant metastases were evaluated through physical examination, laboratory tests, and imaging techniques such as CT or PET-CT. The follow-up period concluded on March 30, 2026. The rate of loss to follow-up was 10.9%. Statistical analysis Statistical analyses were performed utilizing SPSS Statistics 26.0 (IBM Corp, Armonk, NY, USA) and R software (version 4.5.3). Descriptive statistics were employed to summarize baseline characteristics. Categorical variables were compared using the chi-square test or Fisher’s exact test. PFS and OS probabilities were estimated via the Kaplan-Meier method, with group differences assessed using the log-rank test. To identify independent prognostic factors, univariate and multivariate Cox proportional hazards models were constructed to calculate hazard ratios (HRs) with corresponding 95% confidence intervals (CI). Time-dependent receiver operating characteristic (ROC) curve analysis was conducted using the timeROC package in R to evaluate predictive accuracy. Additionally, logistic regression models were established to assess factors associated with radiotherapy toxicity. All tests were two-sided, and a P value of less than 0.05 was considered statistically significant. Results Patient characteristics A total of 126 patients were included in the analysis. Most of them were patients aged 60 and above (80/126, 63.5%). The majority were male (103/126, 81.7%). Eighty-eight patients had a history of smoking (88/126, 69.8%), 87 patients were at stage IIIB (87/126, 69.0%), and the pathological type of 94 patients was squamous cell carcinoma (94/126, 74.6%). The median follow-up period was 53 months, ranging from 10 to 76 months. In addition, the patients were stratified based on the cutoff values (NLR, PNI, ALI, CONUT score and NPS) determined above (Table 1). Table 1Patient characteristicsCharacteristicsCases (No.)Percentage (%)Total126100Sex Female2318.3 Male10381.7Age (years) <604636.5 ≥608063.5Smoking status No3830.2 Yes8869.8ECOG PS 08164.3 14535.7Tumor stage IIIB8769.0 IIIC3931.0Pathological type LUAD3225.4 LUSC9474.6NLR <511188.1 ≥51511.9PNI <452822.2 ≥459877.8CONUT score <410079.4 ≥42620.6ALI <181713.5 ≥1810986.5NPS Group 02822.2 Group 14938.9 Group 24938.9The categorization of NPS is shown in Figure 3. ALI, advanced lung cancer inflammation index; CONUT score, controlled nutritional status score; ECOG PS, Eastern Cooperative Oncology Group performance status; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; PNI, prognostic nutritional index. Patient characteristics The categorization of NPS is shown in Figure 3. ALI, advanced lung cancer inflammation index; CONUT score, controlled nutritional status score; ECOG PS, Eastern Cooperative Oncology Group performance status; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; PNI, prognostic nutritional index. Univariate and multivariate Cox analyses of PFS and OS Univariate analysis revealed that regarding PFS, NLR ≥5 (HR =2.654, 95% CI: 1.406–5.009, P=0.003) and CONUT score ≥4 (HR =4.388, 95% CI: 2.594–7.421, P<0.001) and the high NPS group (HR =3.001, 95% CI: 2.026–4.446, P<0.001) were independent risk factors, while PNI ≥45 (HR =0.328, 95% CI: 0.194–0.555, P<0.001), ALI ≥18 (HR =0.328, 95% CI: 0.194–0.555, P<0.001) were protective factors; Regarding OS, NLR≥5 (HR =3.363, 95% CI: 1.757–6.438, P<0.001), CONUT score ≥4 (HR =5.372, 95% CI: 3.139–9.194, P<0.001), the high NPS group (HR =3.345, 95% CI: 1.769–6.325, P<0.001) were also independent risk factors, while PNI ≥45 (HR =0.274, 95%CI: 0.159–0.472, P<0.001) and ALI ≥18 (HR =0.297, 95% CI: 0.158–0.558, P<0.001) were also protective factors (Table 2). The Kaplan-Meier method was used to analyze the associations between NLR, PNI, ALI, CONUT score and PFS and OS in patients with LA-NSCLC, and the log-rank test was performed. Compared with patients with low NLR, high PNI, high ALI and low CONUT score, patients with high NLR (P=0.002), low PNI (P<0.001), low ALI (P=0.002), and high CONUT score (P<0.001) had significantly worse PFS; for the correlation between NPS and PFS, pairwise comparisons between subgroups further confirmed that the differences in PFS at all levels were statistically significant (group 0 vs. group 1, P=0.01; group 0 vs. group 2, P<0.001; group 1 vs. group 2, P<0.001; Figure 4A-4E). Compared with patients with high NLR, low PNI, low ALI and high CONUT score, patients with low NLR (P=0.002), high PNI (P<0.001), high ALI (P<0.001) and low CONUT score (P<0.001) had longer OS. Similarly, pairwise comparisons between subgroups also confirmed that the differences in OS at all levels were statistically significant (group 0 vs. group 1, P=0.02; group 0 vs. group 2, P<0.001; group 1 vs. group 2, P<0.001; Figure 5A-5E). Given the different construction algorithms and weight distributions of the NLR, PNI, ALI, CONUT score, and NPS, these indicators are considered statistically distinguishable. The results of collinearity diagnosis revealed that there was no multicollinearity among the NLR, PNI, ALI, CONUT score, and NPS (Table S2). On the basis of these findings, we incorporated the aforementioned five indicators in the same multivariate Cox regression model, aiming to mutually correct for the effects of the individual nutrition-inflammation scores, and then to evaluate whether a specific indicator has a relatively independent predictive value for survival outcomes. Regarding PFS, only the CONUT score ≥4 (HR =2.513, 95% CI: 1.336–4.727, P=0.004) and the high NPS group (HR =2.809, 95% CI: 1.517–5.203, P=0.001) were independent risk factors. Regarding OS, only the CONUT score ≥4 (HR =2.908, 95% CI: 1.527–5.538, P=0.001) and the high NPS group (HR =3.345, 95% CI: 1.769–6.325, P<0.001) were independent risk factors (Table 3). These results indicate that CONUT score and NPS are common and independent predictors of PFS and OS. Table 2Univariate analyses of clinical variablesVariablesPFSOSHR95% CIP valueHR95% CIP valueAge (years) <60ReferenceReference ≥601.0100.977–1.0440.551.0180.975–1.0550.30Sex MaleReferenceReference Female0.8850.472–1.6590.700.8710.462–1.6450.67Smoking status Non-smokerReferenceReference Smoker1.1710.681–2.0130.571.0620.613–1.8400.83ECOG PS 0ReferenceReference 11.1110.672–1.8370.681.1370.678–1.9080.63Clinical staging IIIBReferenceReference IIIC1.2690.764–2.1090.361.4780.881–2.4770.14Pathological type LUSCReferenceReference LUAD1.1210.649–1.9370.680.8920.495–1.6050.70NLR <5ReferenceReference ≥52.6541.406–5.0090.003*3.3631.757–6.438<0.001*PNI <45ReferenceReference ≥450.3280.194–0.555<0.001*0.2740.159–0.472<0.001*ALI <18ReferenceReference ≥180.3880.209–0.7190.003*0.2970.158–0.558<0.001*CONUT score <4ReferenceReference ≥44.3882.594–7.421<0.001*5.3723.139–9.194<0.001*NPS Group 0/1ReferenceReference Group 23.0012.026–4.446<0.001*3.3451.769–6.325<0.001*The categorization of NPS is shown in Figure 3. *, P<0.05. ALI, advanced lung cancer inflammation index; CI, confidence interval; CONUT score, controlled nutritional status score; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; OS, overall survival; PFS, progression-free survival; PNI, prognostic nutritional index. Univariate analyses of clinical variables The categorization of NPS is shown in Figure 3. *, P<0.05. ALI, advanced lung cancer inflammation index; CI, confidence interval; CONUT score, controlled nutritional status score; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; OS, overall survival; PFS, progression-free survival; PNI, prognostic nutritional index. Kaplan-Meier survival curves illustrating the impact of NLR (A), PNI (B), ALI (C), CONUT score (D) and NPS (E) on PFS in patients. ALI, advanced lung cancer inflammation index; CONUT score, controlled nutritional status score; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; PFS, progression-free survival; PNI, prognostic nutritional index. Kaplan-Meier survival curves illustrating the impact of NLR (A), PNI (B), ALI (C), CONUT score (D) and NPS (E) on OS in patients. ALI, advanced lung cancer inflammation index; CONUT score, controlled nutritional status score; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; OS, overall survival; PNI, prognostic nutritional index. Table 3Multivariate analyses of clinical variablesVariablesPFSOSHR95% CIP valueHR95% CIP valueNLR <5ReferenceReference ≥51.2780.281–5.8150.751.0700.205–5.5810.94PNI <45ReferenceReference ≥450.6920.360–1.3280.270.6680.347–1.2860.23ALI <18ReferenceReference ≥181.8130.393–8.3690.451.3990.270–7.2530.69CONUT score <4ReferenceReference ≥42.5131.336–4.7270.004*2.9081.527–5.5380.001*NPS Group 0/1ReferenceReference Group 22.8091.517–5.2030.001*3.3451.769–6.325<0.001*The categorization of NPS is shown in Figure 3. *, P<0.05. ALI, advanced lung cancer inflammation index; CI, confidence interval; CONUT score, controlled nutritional status score; HR, hazard ratio; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; OS, overall survival; PFS, progression-free survival; PNI, prognostic nutritional index. Multivariate analyses of clinical variables The categorization of NPS is shown in Figure 3. *, P<0.05. ALI, advanced lung cancer inflammation index; CI, confidence interval; CONUT score, controlled nutritional status score; HR, hazard ratio; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; OS, overall survival; PFS, progression-free survival; PNI, prognostic nutritional index. ROC curve analysis The predictive value of peripheral blood markers and scoring models for prognosis was evaluated by using the ROC curve analysis. For indices yielding an area under the curve (AUC) below 0.5, we took the reverse predicted value for analysis to maintain consistency in direction. When predicting 24-month PFS, the AUC of CONUT score was 0.791 (95% CI: 0.704–0.878), and that of NPS was 0.741 (95% CI: 0.657–0.825), The AUCs of NLR, PNI and ALI were 0.653 (95% CI: 0.541–0.764), 0.711 (95%CI: 0.610–0.811) and 0.679 (95% CI: 0.574–0.783) (Figure 6A). When predicting 36-month PFS, the AUC for the CONUT score was 0.800 (95% CI: 0.724–0.877), while the AUC for the NPS was 0.790 (95% CI: 0.715–0.866). The AUCs for NLR, PNI, and ALI were 0.685 (95% CI: 0.588–0.782), 0.747 (95% CI: 0.660–0.835) and 0.724 (95% CI: 0.632–0.815) (Figure 6B). Furthermore, the AUC for the CONUT score in predicting 3-year OS was 0.823 (95% CI: 0.746–0.900), and for the NPS, it was 0.802 (95% CI: 0.728–0.877). The AUCs for NLR, PNI, and ALI in predicting 3-year OS were 0.719 (95% CI: 0.619–0.820), 0.732 (95% CI: 0.638–0.825) and 0.741 (95% CI: 0.644–0.838) (Figure 6C). Additionally, the AUC for the CONUT score in predicting 5-year OS was 0.817 (95% CI: 0.745–0.889), and for the NPS, it was 0.802 (95% CI: 0.727–0.878). The AUCs for NLR, PNI, and ALI in predicting 3-year OS were 0.677 (95% CI: 0.581–0.773), 0.738 (95% CI: 0.649–0.826) and 0.706 (95% CI: 0.613–0.799) (Figure 6D). We plotted the DeLong test results as a heat map. For details, please refer to Figure S2. ROC curve analysis was performed to assess the sensitivity and specificity of NLR, PNI, ALI, CONUT score, NPS for predicting PFS at 24 and 36 months (A,B) and OS at 3 and 5 years (C,D). ALI, advanced lung cancer inflammation index; AUC, area under the curve; CI, confidence interval; CONUT score, controlled nutritional status score; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; OS, overall survival; PFS, progression-free survival; PNI, prognostic nutritional index; ROC, receiver operating characteristic. Radiotherapy toxicity Among the 126 patients, 58 cases (58/126, 46.0%) experienced radiotherapy-related toxicities of any grade. Among them, we observed 34 cases of grade 2 adverse events, mainly including 19 cases (19/126,15.1%) of grade 2 radiation pneumonitis and 15 cases (15/126, 11.9%) of grade 2 radiation esophagitis. There were only 2 cases (2/126, 1.6%) of grade 3 adverse events, both of which were radiation pneumonitis. However, no grade 4 adverse events were observed. In this study, radiotherapy toxicity endpoints were defined as the occurrence of either grade ≥2 radiation pneumonitis (21/126, 16.7%) or grade ≥2 radiation esophagitis (15/126, 11.9%) during follow-up. The incidence of adverse events varied across different index groups: 5 endpoint events (33.3%) were recorded in the high NLR group, 16 cases (57.1%) in the low PNI group, 15 cases (57.6%) in the high CONUT score group, and 27 cases (55.1%) in the NPS group 2. Univariate analysis showed that a high CONUT score [odds ratio (OR) =5.130, 95% CI: 2.055–12.806, P<0.001] and NPS group 2 (OR =9.273, 95% CI: 3.791–22.683, P<0.001) were risk factors for radiotherapy toxic endpoint events, whereas a high PNI (OR =0.192, 95% CI: 0.079–0.471, P<0.001) was considered a protective factor. In addition, multivariate logistic regression analysis indicated that high CONUT scores (OR =3.194, 95% CI: 1.001–10.207, P=0.049) and NPS group 2 (OR =8.243, 95% CI: 2.946–23.069, P<0.001) were associated with radiotherapy toxicity endpoint events (Table 4). It should be noted that given the limitations of the retrospective analyses, dose-volume indicators, such as lung V5, V20, average lung dose, and esophageal dose parameters, were not included in this study. Consequently, the toxicity analysis results may have been influenced to a certain extent by treatment-related factors and, accordingly, the aforementioned results should be regarded as exploratory. In addition, grade ≥3 radiotherapy toxicity events were found to be extremely rare in this cohort (only 2 cases) and, consequently, the predictive capacity for severe toxicity needs to be interpreted with caution, and a larger sample will be necessary for further verification. Table 4Nutrition-inflammation models and radiotherapy toxicity endpoint eventsVariablesRadiotherapy toxicity endpoint eventsUnivariateMultivariateOR (95% CI)P valueOR (95% CI)P valueNLR <531/111 (27.9%)ReferenceReference ≥55/15 (33.3%)1.290 (0.408–4.078)0.660.259 (0.021–3.241)0.29PNI <4516/28 (57.1%)ReferenceReference ≥4520/98 (20.4%)0.192 (0.079–0.471)<0.001*0.331 (0.088–1.242)0.10CONUT score <421/100 (21.0%)ReferenceReference ≥415/26 (57.7%)5.130 (2.055–12.806)<0.001*3.194 (1.001–10.207)0.049*ALI <187/17 (41.1%)ReferenceReference ≥1829/109 (26.6%)0.222 (0.180–1.488)0.221.977 (0.139–28.068)0.61NPS Group 0/19/77 (11.7%)ReferenceReference Group 227/49 (55.1%)9.273 (3.791–22.683)<0.001*8.243 (2.946–23.069)<0.001*Radiotherapy toxicity endpoints were defined as the occurrence of either grade ≥2 radiation pneumonitis (21/126, 16.7%) or grade ≥2 radiation esophagitis (15/126, 11.9%) during follow-up. The categorization of NPS is shown in Figure 3. *, P<0.05. ALI, advanced lung cancer inflammation index; CI, confidence interval; CONUT score, controlled nutritional status score; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; OR, odds ratio; PNI, prognostic nutritional index. Nutrition-inflammation models and radiotherapy toxicity endpoint events Radiotherapy toxicity endpoints were defined as the occurrence of either grade ≥2 radiation pneumonitis (21/126, 16.7%) or grade ≥2 radiation esophagitis (15/126, 11.9%) during follow-up. The categorization of NPS is shown in Figure 3. *, P<0.05. ALI, advanced lung cancer inflammation index; CI, confidence interval; CONUT score, controlled nutritional status score; NLR, neutrophil-to-lymphocyte ratio; NPS, Naples prognostic score; OR, odds ratio; PNI, prognostic nutritional index. Discussion Establishing a reliable clinical prediction system is crucial for risk stratification, complication management, and formulating individualized treatment plans for patients with LA-NSCLC. To our knowledge, this is the first study to analyze and compare the predictive performance of five nutrition-inflammation scores for survival and radiotherapy toxicity-related events in a cohort of LA-NSCLC patients receiving CRT. NLR was first proposed by Zahorec in 2001 and was initially used as a biomarker of systemic inflammation (22). The PNI, proposed by Japanese scholars Tokio Onodera et al. in 1984, was initially used to evaluate postoperative complications and mortality in patients with gastrointestinal malignancies (23). ALI was proposed by Jafri et al. in 2013 to comprehensively assess the systemic inflammatory status of patients with advanced lung cancer (24). The CONUT score was first proposed by Spanish scholars such as Ignacio de Ulibarri in 2005 and was established in a prospective validation study for hospitalized patients (25). The NPS originated from a retrospective cohort study conducted by a team from the University of Naples, Italy, in 2017 on 562 patients who underwent colorectal cancer surgery. The results showed that the predictive and discriminative abilities of the NPS for OS and disease-free survival (DFS) were superior to those of the PNI and CONUT score. The NPS was confirmed to be a prognostic factor independent of TNM stage (26). Our research found that compared with single-dimensional peripheral blood markers, the multidimensional prognostic assessment tools CONUT score and NPS based on inflammation and nutritional status were common independent predictors of PFS and OS and showed better predictive ability. These two scoring systems also correlate with radiotherapy toxicity. These findings have clinical translational significance and expand the potential utility of nutrition-inflammation scores in multimodal treatment. A key contribution of this study is that nutrition-inflammation scores based on conventional peripheral blood markers, especially the CONUT score and NPS, may provide a simple stratification tool covering both survival prognosis and toxicity risk for patients with LA-NSCLC before CRT and construct a framework integrating the dual dimensions of prognosis and toxicity. The NPS comprises four core indicators: ALB, TC, NLR, and LMR, whereas the CONUT score comprises three indicators: ALB, TC, and TLC. ALB is a core indicator of the nutritional status of the body. A decline in its levels often indicates an imbalance in metabolic homeostasis (27). TC is the core component of the lipid bilayer of the cell membrane. A low cholesterol state can affect the function of membrane receptors, transmembrane signal transduction, and the activity of immune cells, thereby increasing the susceptibility of normal cells to malignant transformation or weakening the body’s immune surveillance ability against cancer cells, ultimately promoting the immune escape of cancer cells (28,29). NLR and LMR include three types of core immunomodulatory cells: neutrophils, lymphocytes, and monocytes. Neutrophils not only inhibit the proliferation of cytotoxic T lymphocytes (CTLs) and weaken the killing activity of natural killer (NK) cells, thereby promoting the immune escape of cancer cells, but also secrete matrix-degrading enzymes that decompose the extracellular matrix and create conditions for cancer cell invasion and distant metastasis. Provides conditions for invasion and distant metastasis of cancer cells (30). Monocytes are recruited to the primary tumor site via chemokines. Driven by signals such as interleukin (IL)-4, IL-10, and transforming growth factor-β (TGF-β), they differentiate into M2 tumor-associated macrophages (TAMs), thereby enhancing the invasion and migration abilities of cancer cells (31). Unlike neutrophils and monocytes, lymphocytes exert antitumor immune effects through immune surveillance, initiation of adaptive immune responses, and regulation of the inflammatory microenvironment (32). The mechanisms jointly determine the core role of the three types of cells in the occurrence, progression, and metastasis of tumors, and make the NLR and LMR derived from them important reference indicators for the prognosis assessment of malignant tumors. The core biological significance of the CONUT score and NPS lies in systematically reflecting the dynamic correlation among nutritional metabolism, inflammatory responses, and immunosuppression in the tumor microenvironment. This multidimensional comprehensive assessment model overcomes the limitations of single-dimensional markers and provides a biological rationale for their prognostic value. This study also identified a clear “protection-risk” bipolar pattern: in the univariate analysis, PNI and ALI were protective factors for OS and PFS, while NLR, CONUT score and NPS were risk factors for OS and PFS. Among the protective factors, multiple large-sample studies and meta-analyses have confirmed that a high PNI is usually associated with better PFS and OS in patients with lung cancer (18,33,34). A multicenter retrospective study involving 1,772 patients with lung cancer demonstrated that low ALI (HR =1.30, 95% CI: 1.13–1.49, P<0.001) was associated with a higher risk of lung cancer mortality (19). High ALI usually indicates a better nutritional status (high BMI, high ALB), accompanied by a lower inflammatory load (low NLR), which can be understood as a ratio protective factor of systemic nutritional reserves divided by systemic inflammatory load. Among these risk factors, a high NLR simultaneously reflects increased systemic inflammatory burden driven by neutrophil- and lymphocyte-mediated adaptive immune exhaustion. A recent study has shown that the increase in NLR during radiotherapy and chemotherapy is mainly driven by a sudden drop in TLC, whereas neutrophils remain relatively stable, suggesting that the direct cytotoxic effect of radiotherapy and chemotherapy on lymphocytes exacerbates the poor prognosis caused by a high NLR (35). For the CONUT score, Peng et al. included 11 studies involving 3,029 patients. The results showed that a high CONUT score was positively correlated with poor OS (HR =1.63, 95% CI: 1.40–1.88, P<0.01), and that a high preoperative CONUT score was an independent predictor of postoperative complications in NSCLC (OR =1.58, 95% CI: 1.21–2.06, P=0.001) (36). Xie et al. included 15 high-quality studies involving 4,973 Asian patients with NSCLC in a meta-analysis. The results showed that a high CONUT score was associated with poor OS (HR =1.84, 95% CI: 1.55–2.18, P<0.001) (12). Wang et al. included seven lung cancer studies in their meta-analysis. The results showed that high NPS was a significant predictor of OS (HR =3.21, 95% CI: 2.27–4.54, P<0.001) (37). Yang et al. further included 12 studies involving 3,089 patients with lung cancer in a meta-analysis and found that a high NPS was significantly associated with poor OS (HR =2.82, 95% CI: 1.88–4.25, P<0.001) and PFS (HR =2.75, 95% CI: 1.89–4.01, P<0.001) (38). Our results are also consistent with those of the abovementioned meta-analyses. Together, evidence from these meta-analyses and the present study suggests that the CONUT score and NPS can serve as robust quantitative bases for prognostic risk. The clinical significance of these scores lies in their ability to predict the survival of patients with LA-NSCLC receiving CRT independently. It is an economical, efficient, and easily accessible bedside risk-stratification tool that can effectively identify high-risk populations and provide preventive nutritional support. In the era of immunotherapy, while CRT offers significant survival benefits to patients with LA-NSCLC, it is inevitably accompanied by radiotherapy toxicity, which can lead to treatment interruptions or even more severe harm (39). Our research found that PNI, CONUT score, and NPS were associated with the occurrence of radiotherapy toxicity-related events to some extent. This reflects the support of nutritional reserves, cholesterol metabolism, and immune response to the normal tissue repair capacity, while radiotherapy toxicity during radiotherapy and chemotherapy mainly depends on the “dynamic game” between radiation damage and normal tissue repair. A low ALB level usually indicates hypotonic malnutrition and reduced protein synthesis, thereby delaying the repair of sublethal damage in normal tissues. Lymphocytes are highly radiation-sensitive cells. Patients with a low TLC are more prone to severe lymphopenia during radiotherapy and chemotherapy, thereby increasing the risk of infection and the probability of treatment interruption (40). Previously, cholesterol was mainly regarded as a passive structural component of cell membranes; however, recent cross-scale mechanistic studies have revealed its key regulatory role in radiation-induced damage to normal tissues. At the membrane signaling level, the DNA-dependent protein kinase complex (DNA-PK) is located in the cholesterol-rich lipid raft region. After irradiation, this complex showed a differential phosphorylation pattern in lipid rafts (41). Artificial alteration of cholesterol distribution in lipid rafts can lead to DNA double-strand breaks and insufficient activation of DNA-PK (42). At the metabolic reprogramming level, radiotherapy enhances cholesterol synthesis by upregulating HMGCR, leading to chelation of STING to the endoplasmic reticulum and inhibition of cGAS-STING pathway activation (43). More importantly, cholesterol can directly promote chromatin compression, and the accumulation of cholesterol in the nucleus weakens chromatin accessibility by inhibiting BRG1 activity and impairing DNA damage sensing and checkpoint activation. As a consequence of these repair defects, the cells carry unrepaired DNA into mitosis, forming cytoplasmic chromatin fragments and activating the cGAS-STING cascade, ultimately inducing cellular senescence (44). The senescent secretory phenotype (SASP) released by senescent cells is rich in factors such as IL-1 and tumor necrosis factor (TNF) and has been confirmed to drive the occurrence of radiation-induced pulmonary fibrosis (45). The SASP also impairs the cell burial function of macrophages, creating a positive feedback loop for the accumulation of senescent cells and amplification of inflammation, ultimately leading to irreversible repair disorders. These biological mechanisms suggest that cholesterol metabolism may serve as an upstream node for intervention. Interventions targeting cholesterol metabolism in patients with a high CONUT score/NPS (such as nutritional support to adjust lipid metabolism and the combination of statins) may reduce radiotherapy toxicity. This study has several limitations. First, this was a retrospective study conducted in a single center with a limited sample size, which may have led to bias in the data analysis. Moreover, an external validation cohort was lacking. The optimal cutoff values for peripheral blood markers and scoring systems mainly refer to previous literature, and their universality needs to be verified by multicenter prospective studies. Second, although we used strict inclusion and exclusion criteria, peripheral blood indicators such as ALB, TC, neutrophils, and lymphocytes may have been influenced by multiple factors, thereby affecting the treatment outcomes of these patients. Finally, owing to the limited number of cases corresponding to radiotherapy toxicity-related events, this may reduce estimation accuracy. Although this limits the precise inference of the predictive effect to a certain extent, it does not negate its potential predictive value. Previous studies have confirmed that multiple radiation dosimetric parameters [such as V20, V5, and mean lung dose (MLD)] are predictive factors for radiotherapy toxicity (46,47). The association of CONUT score and NPS with grade ≥2 radioactive lung/esophageal toxicity in this study may be partly explained by the differences in unmeasured dose parameters rather than as an independent effect of nutrition–inflammatory status. Consequently, the toxicity prediction conclusions of this study should be viewed with caution and interpreted as causal inferences. Based on the findings of this study, we suggest that subsequent research should focus on the following directions: conduct prospective, multicenter cohort studies to verify the predictive efficacy of NPS and CONUT score; integrate the nutrition-inflammation score with the dose parameters of radiotherapy to construct a joint prediction model; explore the interaction between the two; and design interventional studies to evaluate whether supportive treatment for patients with high NPS and CONUT scores can improve treatment tolerance and survival outcomes. The advancement of the above studies will help to achieve precise stratification of LA-NSCLC prior to treatment, with a view to translating into personalized clinical decisions. Conclusions In this retrospective study cohort, we identified pre-treatment CONUT score and NPS as predictive factors for PFS and OS in LA-NSCLC patients receiving CRT, and established that these indices had good predictive value. The pre-treatment CONUT score and NPS were correlated with grade ≥2 radiation pneumonitis and esophagitis. However, these findings will need to be confirmed by performing prospective multi-center studies to evaluate their potential clinical translation. Supplementary The article’s supplementary files as Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1466/prf